Prior to 2014, I contributed to the Toss games modeling system
Toss: General Game Playing and Hybrid Modelling. Paper highlight: First-Order Logic with Counting for General Game Playing.
February 2012 - March 2013, I created slides on FP topics in OCaml; now converted to a textbook and modernized!
HTML version: https://lukstafi.github.io/curious-ocaml/new_book.htmlPDF version: https://github.com/lukstafi/curious-ocaml/blob/main/pdfs/new_book.pdf
Original slides below. I used slide decks (1) - (8) in a 14-week course.
(1) Logic, (2) Algebra (exercises), (3) Computation (exercises), (4) Functions (exercises), (5) Polymorphism & ADTs (exercises), (6) Folding and Backtracking (exercises), (7) Laziness (exercises), (8) Monads (exercises), (9) Compiler (exercises), (10) FRP (exercises), (11) The Expression Problem (exercises); (at a glance except 11).
In 2015, I defended my PhD thesis on a system InvarGenT for inferring the properties of programs
Introduction slides -- good to check first, thesis GADTs for Reconstruction of Invariants and Postconditions.
Between 2015-2021 I worked at Google
I worked on knowledge representation and semantic parsing (in C++), then briefly on deep reinforcement learning (Python/JAX).OCANNL: a from-scratch deep learning framework fully in OCaml
OCANNL is my main project currently. Highlights: slides Shapes and Einsum, migration guide from PyTorch to OCANNL.
Since April 2024, my work on OCANNL is sponsored by Ahrefs! Visit the Ahrefs website.